Expression of potential molecular markers in renal cell carcinoma: impact on clinicopathological outcomes in patients undergoing radical nephrectomy

Authors


Hideaki Miyake, Division of Urology, Kobe University Graduate School of Medicine, 7-5-1 Kusunoki-cho, Chuo-ku, Kobe 650-0017, Japan.
e-mail: hideakimiyake@hotmail.com

Abstract

OBJECTIVES

To evaluate the expression levels of several potential molecular markers in renal cell carcinoma (RCC), to clarify the significance of these markers as prognostic predictors in patients undergoing radical nephrectomy (RN).

PATIENTS AND METHODS

The study included 153 patients with clinically organ-confined RCC undergoing RN. Expression levels of 12 proteins, including Aurora-A, Bcl-2, Bcl-xL, clusterin, heat-shock protein 27 (HSP27), HSP70, HSP90, Ki-67, matrix metalloproteinase (MMP)-2 and -9, p53 and vascular endothelial growth factor, in RN specimens obtained from these 153 patients were measured by immunohistochemical staining.

RESULTS

Of the 12 markers, clusterin, HSP27, Ki-67, MMP-2 and -9 expression were significantly associated with several conventional prognostic factors. Univariate analysis also identified these five markers as significant predictors of disease recurrence, while mode of presentation, pathological stage, grade and microvascular invasion were also significant. Of these significant factors, Ki-67 expression, pathological stage and microvascular invasion appeared to be independently related to disease recurrence. Furthermore, there were significant differences in recurrence-free survival according to positive numbers of these three independent factors, i.e. disease recurred in four of 56 patients who were negative for risk factors (7%), 17 of 71 positive for one risk factor (24%), and 20 of 26 positive for two or three risk factors (77%).

CONCLUSIONS

Combined evaluation of Ki-67 expression, pathological stage and microvascular invasion would be particularly useful for further refinement of the system for predicting the outcome after RN for patients with RCC.

Abbreviations
RN

radical nephrectomy

HSP

heat-shock protein

MMP

matrix metalloproteinase

VEGF

vascular endothelial growth factor

MVI

microvascular invasion

RFS

recurrence-free survival.

INTRODUCTION

Radical nephrectomy (RN) has been regarded as an effective therapy for patients with clinically localized RCC, but 20–30% of these patients have a local and/or distant disease recurrence [1]. Thus intensive efforts have been made to identify variables that precisely predict the outcome of RN, and which are of great utility in planning both postoperative therapy and the follow-up schedule in an individual patient.

To date, various factors, including pathological stage, tumour grade, histological subtype, microvascular invasion (MVI) and inflammatory response, have been shown to be significantly associated with treatment failure after RN [2,3]. Furthermore, several investigators have shown the usefulness of nomograms in calculating the probability of disease recurrence after surgical resection of localized RCC [4,5]. However, RCC has been shown to be characterized by unique biological features and a heterogeneous genetic background. Indeed, RCCs with similar clinicopathological factors can show different biological behaviour and result in different clinical outcomes [1]. These findings suggest the limitations for predicting outcomes after RN in patients with localized RCC using conventional variables alone.

In an attempt to provide more reliable prognostic information in patients with RCC, several studies examined the value of a wide variety of molecular markers, and some of these molecules were shown to be significantly related to treatment failure [2,3,6–10]. For example, Kim et al.[6] reported that carbonic anhydrase-9, PTEN, vimentin and p53 could be used as independent predictors of disease-specific survival in patients with metastatic RCC who had RN before immunotherapy. However, there have been a few studies evaluating the prognostic significance of multiple molecular markers in localized rather than metastatic RCC. Accordingly, in the present study, we evaluated the expression levels of 12 potential molecular markers, including Aurora-A, Bcl-2, Bcl-xL, clusterin, heat-shock protein 27 (HSP27), HSP70, HSP90, Ki-67, matrix metalloproteinase (MMP)-2 and -9, p53 and vascular endothelial growth factor (VEGF), in RN specimens with immunohistochemical staining, and analysed these outcomes according to several clinicopathological variables.

PATIENTS AND METHODS

This study included 153 patients who were diagnosed as having clinically organ-confined RCC, and subsequently had RN between 1998 and 2004. The specimens were examined for pathology by one pathologist according to the 2002 American Joint Committee on Cancer TNM classification system. Information on clinicopathological characteristics of included patients was retrieved from their medical records. The median (range) duration of follow-up after RN of this series was 69 (39–117) months. Generally, all patients were followed by laboratory and radiological evaluations, and a physical examination, every 6 months to monitor recurrence and metastasis. In the absence of a relapse of RCC by 5 years after surgery, the interval between re-examinations was increased. Informed consent for this study was obtained from all of these patients, and the study design was approved by the Research Ethics Committee of our institution.

The RN specimens were stained immunohistochemically as previously described [11]. Briefly, sections from formaldehyde-fixed, paraffin-embedded tissue from the 153 specimens were deparaffinized by xylene and rehydrated in decreasing concentrations of ethanol. After blocking endogenous peroxidase with 3% hydrogen peroxidase in methanol, sections were boiled on 0.01 m citrate buffer for 10 min and incubated with 5% normal blocking serum in Tris-buffered saline for 20 min The sections were then incubated with the following antibodies: antihuman Aurora-A rabbit polyclonal antibody (Abcam, Cambridge, UK), antihuman Bcl-2 mouse monoclonal antibody (Dako, Carpinteria, CA, USA), antihuman Bcl-xL mouse monoclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA), antihuman clusterin rabbit polyclonal antibody (Santa Cruz Biotechnology), antihuman HSP27 mouse monoclonal antibodies (Novocastra Laboratories, Newcastle, UK), antihuman HSP70 mouse monoclonal antibodies (Novocastra), antihuman HSP90 mouse monoclonal antibodies (Novocastra), antihuman Ki-67 mouse monoclonal antibody (Dako), antihuman MMP-2 mouse monoclonal antibody (Daiichi Fine Chemical, Takaoka, Japan), antihuman MMP-9 mouse monoclonal antibody (Daiichi Fine Chemical), antihuman p53 mouse monoclonal antibody (Novocastra) and antihuman VEGF rabbit polyclonal antibody (Santa Cruz). The sections were then incubated with biotinylated goat antimouse or rabbit IgG (Vector Laboratories, Burlingame, CA, USA). After incubation in an avidin–biotin peroxidase complex for 30 min, the samples were exposed to diaminobenzidine tetrahydrochloride solution and counterstained with haematoxylin.

Staining results were interpreted by two independent observers who were unaware of the clinicopathological data. For Ki-67 and p53 analyses, only nuclear staining was considered, and strong expression of Ki-67 and p53 was defined as the proportion of positively stained tumour cells >5% and >20%, respectively, as previously reported [12]. The highest intensity of immunohistochemical staining for the remaining 10 proteins was visually scored from several fields of each section and classified as negative, weak, moderate and strong. According to the previous study [13], either moderate or strong staining intensity in >10% of tumour cells was considered as strong expression.

The chi-squared test was used to analyse the association between several clinicopathological factors and expression levels of molecular markers. The recurrence-free survival (RFS) rates were calculated by the Kaplan-Meier method, and differences were determined using the log-rank test. The prognostic significance of certain factors was assessed by the Cox proportional hazards regression model. In all tests, P < 0.05 was considered to indicate significance.

RESULTS

The characteristics of the 153 patients included in the study are shown in Table 1. The association of the clinicopathological factors and expression levels of the 12 potential molecular markers in RN specimens was then analysed. As shown in Table 2, clusterin, HSP27, Ki-67, MMP-2 and MMP-9 expression were significantly associated with several conventional prognostic variables, whereas expression levels of the remaining seven proteins showed significant correlation with either no or limited prognostic variables.

Table 1.  Characteristics of 153 patients with RCC who had RN
VariableMedian (range) or n (%)
Age, years 60.5 (30–87)
Male/female 93 (60.8)/60 (39.2)
Mode of presentation
 Incidental109 (71.2)
 Symptomatic 44 (28.8)
Pathological stage
 pT1101 (66.0)
 pT2 29 (18.9)
 pT3 20 (13.1)
 pT4  3 (2.0)
Grade
 1 80 (52.3)
 2 67 (43.8)
 3  6 (3.9)
MVI
 Negative 88 (57.5)
 Positive 65 (42.5)
Histological subtype
 Clear cell132 (86.3)
 Not clear cell 21 (13.7)
Table 2.  Correlation of expression levels of molecular markers in RN specimens with several clinicopathological factors
FactorMarkerTotal patients
Aurora-ABcl-2Bcl-xLClusterinHSP27HSP70HSP90Ki-67MMP-2MMP-9VEGFp53
  1. Values are expressed as the number of patients with strong expression of each molecule.

Age, years
 <6021575642375553423936 5926 76
 ≥6020565438335350464341 5825 77
 P 0.82 0.75 0.62 0.46 0.52 0.63 0.53 0.58 0.57 0.53  0.74 0.82 
Gender
 Male28676448426461504744 7032 93
 Female13464632284442383533 4719 60
 P 0.25 0.96 0.29 0.84 0.86 0.55 0.57 0.24 0.35 0.35  0.66 0.73 
Mode of presentation
 Incidental28857665567574757062 8734109
 Symptomatic13283415143329131215 3017 44
 P 0.63 0.068 0.35 0.004 0.028 0.45 0.81<0.001<0.001 0.011  0.12 0.38 
Pathological stage
 pT131727161547269696661 7535101
 ≥ pT210413919163634191616 4216 52
 P 0.13 0.31 0.54 0.005 0.008 0.79 0.71<0.001<0.001<0.001  0.37 0.63 
Grade
 127605550435351555248 6232 80
 2 or 314535530275552333029 5519 73
 P 0.042 0.74 0.36 0.008 0.038 0.22 0.97 0.003 0.003 0.012  0.75 0.067 
MVI
 −ve23646355476263676359 6728 88
 +ve18494725234640211918 5023 65
 P 0.83 0.71 0.92 0.003 0.027 0.97 0.19<0.001<0.001<0.001  0.91 0.64 
Histological subtype
 Clear cell3696967063938976706610045132
 Not clear cell 5171410 715141212 11 17 6 21
 P 0.74 0.43 0.57 0.64 0.22 0.93 0.95 0.97 0.73 0.84  0.60 0.62 

During the observation period of the study, 41 of the 153 patients (26.8%) developed disease recurrence, and the 1-, 3- and 5-year RFS rates were 94.8%, 79.0% and 74.2%, respectively (Fig. 1). By univariate analysis using the Cox proportional hazards regression model, expression levels of clusterin, HSP27, Ki-67, MMP-2 and MMP-9 were identified as significant factors associated with RFS, while mode of presentation, pathological stage, grade and MVI were also significant among several clinicopathological factors examined (Table 3). Furthermore, multivariate analysis of nine significant factors determined by the univariate analysis was used to evaluate the predictive value for disease recurrence, and showed that Ki-67 expression level, pathological stage and MVI were independently associated with disease recurrence irrespective of other factors included (Table 3). RFS curves according to Ki-67 expression level, pathological stage and MVI are shown in Fig. 2. There were significant differences in RFS for all these three factors.

Figure 1.

RFS of 153 patients with clinically localized RCC after RN.

Table 3.  Univariate and multivariate analyses of association between several variables with RFS in 153 patients with RCC who had RN
VariableN patients (n with disease recurrence)P
UnivariateMultivariate
Age, years (<60 vs ≥60)76 (19) vs 77 (22)0.660
Gender (male vs female)93 (27) vs 60 (14)0.420
Mode of presentation (incidental vs symptomatic) 109 (21) vs 44 (20)0.0360.230
Pathological stage (pT1 vs ≥pT2) 101 (19) vs 52 (22)0.0020.042
Grade (1 vs 2 or 3)80 (14) vs 73 (27)0.0180.330
MVI (negative vs positive)88 (12) vs 65 (29)<0.0010.018
Histological subtype (clear cell vs not)132 (36) vs 21 (5)0.630
Weak expression vs strong expression for:
Aurora-A41 (12) vs 112 (29)0.140
Bcl-2 113 (30) vs 40 (11)0.330
Bcl-xL110 (31) vs 43 (10)0.480
Clusterin80 (14) vs 73 (27)0.0300.180
HSP2770 (12) vs 83 (29)0.0370.370
HSP70 108 (29) vs 45 (12)0.260
HSP90 103 (27) vs 50 (14)0.550
Ki-6788 (17) vs 65 (24)<0.0010.037
MMP-282 (17) vs 71 (24)0.0190.089
MMP-977 (14) vs 76 (27)0.0290.360
p53117 (31) vs 36 (10)0.310
VEGF51 (17) vs 102 (24)0.280
Figure 2.

RFS of patients with clinically localized RCC after RN, according to expression level of Ki-67 (A), pathological stage (B) and MVI (C).

To more precisely characterize the prognostic features after RN we categorized patients according to positive numbers of three independent risk factors for disease recurrence, including Ki-67 expression level, pathological stage and MVI. Disease recurrence occurred in four of 56 patients who had no risk factor (71%), 17 of 71 positive for one risk factor (24%), and 20 of 26 positive for two or three risk factors (77%). As shown in Fig. 3, there were significant differences in RFS among these three groups.

Figure 3.

RFS of patients with clinically localized RCC after RN according to the number of independent risk factors for disease recurrence, including strong Ki-67 expression, advanced pathological stage (i.e. ≥ pT2) and MVI.

DISCUSSION

A precise prediction of postoperative prognosis in patients undergoing surgical resection for malignant tumours, using a worldwide consensus, would be important for physicians to make decisions about the follow-up schedule and additional treatment, which is particularly true in patients with RCC because of its highly resistant phenotype to conventional non-surgical treatments [1]. Historically, staging according to the TNM classification system has been the standard for stratifying patients into risk groups for prognosis; however, recent studies indicate that stage alone is not reliable for predicting disease recurrence in patients with localized RCC [2,3,14]. Consequently, several molecular markers, in addition to conventional variables, have been investigated as prognostic indicators in RCC. Although some of these systems were of some use in predicting postoperative clinical outcomes in patients with RCC [2,3,6–10], there have been no systems widely introduced into clinical practice, due to several limitations. In the present study therefore we analysed the values of conventional clinicopathological prognostic factors and several potential molecular markers for predicting the clinical outcome in 153 patients undergoing RN for clinically localized RCC.

There were limitations to this study; a sample size of 153 patients in such a comparatively common disease like RCC is not large enough. In addition, although the study included 12 proteins as potential molecular markers for RCC, these 12 had not been selected based on scientifically objective criteria. Therefore, there might be other molecules more closely associated with the prognostic outcome in RCC than these 12 molecules. However, most previous studies focused on the significance of either one or a few molecular markers in each series [2,3,6–10], and some of them used tissue microarrays, which do not always provide representative information in an individual case [6,7]. Collectively, these findings suggest the advantages of the present study, simultaneously investigating 12 molecular markers using immunohistochemical staining of the RCC specimen from each patient.

Expression of the 12 molecular markers investigated in this study was detected in most of RCC tissues. However, there were varied patterns in the relation between the expression level of each molecule and the conventional prognostic variables, i.e. expression levels of clusterin, HSP27, Ki-67, MMP-2 and MMP-9 had a significant effect on these variables, whereas expression levels of the remaining seven proteins showed a significant association with either no or limited prognostic factors. However, the significance of these seven proteins in RCC progression cannot be completely denied based on the present findings. For example, Aurora-A has been shown to be widely expressed in various types of malignant disease, irrespective of the degree of disease progression, and thus regarded as being involved in the initiation rather than the progression of malignant diseases [15]. There was strong expression of Aurora-A protein in 73.2% of RCC specimens evaluated in the present series. Accordingly, it would be necessary to interpret the findings of immunohistochemical staining considering the functional characteristics of evaluated molecular markers.

The effect of conventional prognostic variables and potential molecular markers on RFS after RN were subsequently compared. Univariate analysis showed that mode of presentation, pathological stage, grade and MVI, expression levels of clusterin, HSP27, Ki-67, MMP-2 and -9 were significantly associated with disease recurrence; however, only the expression level of Ki-67, pathological stage and MVI were independent prognostic predictors by multivariate analysis. Pathological stage and MVI are well established prognostic factors after RN [2,5,16], and the usefulness of the Ki-67 expression level as a prognostic marker has been also confirmed in previous studies [6–9]. Therefore, it is interesting to evaluate the ability to predict disease recurrence after RN on combining these three independent risk factors, as using an approach similar to that in the present study, Dall’Oglio et al.[17] reported that risk assessment and stratification based on MVI, tumour size and Fuhrman grade in RN specimens might allow better individualization of follow-up and treatment schedules for patients with RCC. In the present study, when the 153 patients were classified into the three groups of no risk factors, positive for one and positive for two or three, there were significant differences in RFS among these groups. These findings suggest that consideration of the three main risk factors identified by multivariate analysis (i.e. Ki-67 expression level, pathological stage and MVI) might contribute to developing a novel system that can more precisely predict disease recurrence after RN.

In the present series, we investigated the significance of several conventional prognostic variables and expression levels of several potential molecular markers in RN specimens simultaneously, as predictors of disease recurrence for clinically localized RCC, and expression levels of Ki-67, pathological stage and MVI were identified as independent prognostic predictors after RN. Moreover, it has been suggested that the utility of the combined use of these three factors could contribute to further refinement of this predictive system. However, a prospective study investigating additional potential molecular markers involved in the progression of RCC will be necessary before a definitive conclusion can be drawn.

CONFLICT OF INTEREST

None declared.

Ancillary